Causality
Three weeks to causally competent.
Learn why modern causal estimators work by watching them move. No proofs required to start; all of the proofs are here when you want them.
Written for trial statisticians and epidemiologists who know regression, Cox and Kaplan–Meier, and want to see what AIPW and TMLE are actually doing.
What you will be able to do
Write the estimand first, then say which assumptions turn it into something the data can answer.
Say what a one-step or TMLE correction removes, what it leaves, and when its interval is honest.
Choose between a hazard ratio, a survival difference and RMST as answers to different questions.
Fix a biased estimate in one step
A plug-in estimate misses the truth. Take one step along the tangent, the influence function, and see how close it lands. Drag the slider, take the step again, and watch what is left over. By the middle of the course you will know exactly why.
Or just a 3-question identification check
Answer all three and the first lesson is marked skimmed, so the route starts you at the lesson after it.
The route
One recurring cohort of 100 patients ties the course together, with clearly marked teaching worlds when a new idea needs a different setting. Every lesson opens directly; prerequisites are advice, not gates.
Take it to your SAP
Describe a randomized trial and get back the estimand, draft SAP text for a covariate-adjusted primary analysis, the patients it saves, simulated operating characteristics, the questions a reviewer will ask, and R code that runs.